Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 1 | /* vi: set sw=4 ts=4: */ |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 2 | /* |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 3 | * sed.c - very minimalist version of sed |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 4 | * |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 5 | * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley |
| 6 | * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org> |
Matt Kraai | 5ed78ad | 2002-01-03 21:12:34 +0000 | [diff] [blame] | 7 | * Copyright (C) 2002 Matt Kraai |
Denis Vlasenko | 0beaff8 | 2007-09-21 13:16:32 +0000 | [diff] [blame] | 8 | * Copyright (C) 2003 by Glenn McGrath |
Rob Landley | 25d8239 | 2004-04-01 09:23:30 +0000 | [diff] [blame] | 9 | * Copyright (C) 2003,2004 by Rob Landley <rob@landley.net> |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 10 | * |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 11 | * MAINTAINER: Rob Landley <rob@landley.net> |
Bernhard Reutner-Fischer | b7f3973 | 2006-03-01 20:14:16 +0000 | [diff] [blame] | 12 | * |
Rob Landley | e9a7a62 | 2006-09-22 02:52:41 +0000 | [diff] [blame] | 13 | * Licensed under GPL version 2, see file LICENSE in this tarball for details. |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 14 | */ |
| 15 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 16 | /* Code overview. |
| 17 | |
| 18 | Files are laid out to avoid unnecessary function declarations. So for |
| 19 | example, every function add_cmd calls occurs before add_cmd in this file. |
| 20 | |
| 21 | add_cmd() is called on each line of sed command text (from a file or from |
| 22 | the command line). It calls get_address() and parse_cmd_args(). The |
| 23 | resulting sed_cmd_t structures are appended to a linked list |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 24 | (G.sed_cmd_head/G.sed_cmd_tail). |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 25 | |
Denis Vlasenko | defc1ea | 2008-06-27 02:52:20 +0000 | [diff] [blame] | 26 | add_input_file() adds a FILE* to the list of input files. We need to |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 27 | know all input sources ahead of time to find the last line for the $ match. |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 28 | |
| 29 | process_files() does actual sedding, reading data lines from each input FILE * |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 30 | (which could be stdin) and applying the sed command list (sed_cmd_head) to |
| 31 | each of the resulting lines. |
| 32 | |
| 33 | sed_main() is where external code calls into this, with a command line. |
| 34 | */ |
| 35 | |
| 36 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 37 | /* |
| 38 | Supported features and commands in this version of sed: |
| 39 | |
| 40 | - comments ('#') |
Mark Whitley | 94074a9 | 2000-07-14 00:00:15 +0000 | [diff] [blame] | 41 | - address matching: num|/matchstr/[,num|/matchstr/|$]command |
| 42 | - commands: (p)rint, (d)elete, (s)ubstitue (with g & I flags) |
| 43 | - edit commands: (a)ppend, (i)nsert, (c)hange |
Mark Whitley | 1f3b9f2 | 2001-05-11 22:27:13 +0000 | [diff] [blame] | 44 | - file commands: (r)ead |
Rob Landley | c63fe91 | 2005-10-30 10:08:13 +0000 | [diff] [blame] | 45 | - backreferences in substitution expressions (\0, \1, \2...\9) |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 46 | - grouped commands: {cmd1;cmd2} |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 47 | - transliteration (y/source-chars/dest-chars/) |
| 48 | - pattern space hold space storing / swapping (g, h, x) |
Rob Landley | 93850a5 | 2005-05-18 06:34:37 +0000 | [diff] [blame] | 49 | - labels / branching (: label, b, t, T) |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 50 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 51 | (Note: Specifying an address (range) to match is *optional*; commands |
| 52 | default to the whole pattern space if no specific address match was |
| 53 | requested.) |
| 54 | |
Glenn L McGrath | 2570b43 | 2003-09-16 05:25:43 +0000 | [diff] [blame] | 55 | Todo: |
Glenn L McGrath | 2570b43 | 2003-09-16 05:25:43 +0000 | [diff] [blame] | 56 | - Create a wrapper around regex to make libc's regex conform with sed |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 57 | |
Glenn L McGrath | d5eadea | 2003-03-09 02:39:29 +0000 | [diff] [blame] | 58 | Reference http://www.opengroup.org/onlinepubs/007904975/utilities/sed.html |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 59 | */ |
| 60 | |
Denis Vlasenko | b6adbf1 | 2007-05-26 19:00:18 +0000 | [diff] [blame] | 61 | #include "libbb.h" |
"Vladimir N. Oleynik" | 23f62fc | 2005-09-14 16:59:11 +0000 | [diff] [blame] | 62 | #include "xregex.h" |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 63 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 64 | /* Each sed command turns into one of these structures. */ |
Glenn L McGrath | e7a8bc9 | 2003-03-09 10:23:57 +0000 | [diff] [blame] | 65 | typedef struct sed_cmd_s { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 66 | /* Ordered by alignment requirements: currently 36 bytes on x86 */ |
Denis Vlasenko | 34c4e5f | 2007-01-30 22:25:16 +0000 | [diff] [blame] | 67 | struct sed_cmd_s *next; /* Next command (linked list, NULL terminated) */ |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 68 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 69 | /* address storage */ |
| 70 | regex_t *beg_match; /* sed -e '/match/cmd' */ |
| 71 | regex_t *end_match; /* sed -e '/match/,/end_match/cmd' */ |
| 72 | regex_t *sub_match; /* For 's/sub_match/string/' */ |
| 73 | int beg_line; /* 'sed 1p' 0 == apply commands to all lines */ |
| 74 | int end_line; /* 'sed 1,3p' 0 == one line only. -1 = last line ($) */ |
Bernhard Reutner-Fischer | b7f3973 | 2006-03-01 20:14:16 +0000 | [diff] [blame] | 75 | |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 76 | FILE *sw_file; /* File (sw) command writes to, -1 for none. */ |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 77 | char *string; /* Data string for (saicytb) commands. */ |
Glenn L McGrath | fc4cb4d | 2003-04-12 16:10:42 +0000 | [diff] [blame] | 78 | |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 79 | unsigned which_match; /* (s) Which match to replace (0 for all) */ |
Glenn L McGrath | fc4cb4d | 2003-04-12 16:10:42 +0000 | [diff] [blame] | 80 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 81 | /* Bitfields (gcc won't group them if we don't) */ |
Denis Vlasenko | 34c4e5f | 2007-01-30 22:25:16 +0000 | [diff] [blame] | 82 | unsigned invert:1; /* the '!' after the address */ |
| 83 | unsigned in_match:1; /* Next line also included in match? */ |
| 84 | unsigned sub_p:1; /* (s) print option */ |
Denis Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 85 | |
Denis Vlasenko | 34c4e5f | 2007-01-30 22:25:16 +0000 | [diff] [blame] | 86 | char sw_last_char; /* Last line written by (sw) had no '\n' */ |
Glenn L McGrath | c18ce37 | 2003-09-13 06:57:39 +0000 | [diff] [blame] | 87 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 88 | /* GENERAL FIELDS */ |
| 89 | char cmd; /* The command char: abcdDgGhHilnNpPqrstwxy:={} */ |
Glenn L McGrath | e7a8bc9 | 2003-03-09 10:23:57 +0000 | [diff] [blame] | 90 | } sed_cmd_t; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 91 | |
Denis Vlasenko | 6ca409e | 2007-08-12 20:58:27 +0000 | [diff] [blame] | 92 | static const char semicolon_whitespace[] ALIGN1 = "; \n\r\t\v"; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 93 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 94 | struct globals { |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 95 | /* options */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 96 | int be_quiet, regex_type; |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 97 | FILE *nonstdout; |
| 98 | char *outname, *hold_space; |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 99 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 100 | /* List of input files */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 101 | int input_file_count, current_input_file; |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 102 | FILE **input_file_list; |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 103 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 104 | regmatch_t regmatch[10]; |
| 105 | regex_t *previous_regex_ptr; |
Denis Vlasenko | 9213a9e | 2006-09-17 16:28:10 +0000 | [diff] [blame] | 106 | |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 107 | /* linked list of sed commands */ |
| 108 | sed_cmd_t sed_cmd_head, *sed_cmd_tail; |
| 109 | |
| 110 | /* Linked list of append lines */ |
| 111 | llist_t *append_head; |
| 112 | |
| 113 | char *add_cmd_line; |
| 114 | |
| 115 | struct pipeline { |
| 116 | char *buf; /* Space to hold string */ |
| 117 | int idx; /* Space used */ |
| 118 | int len; /* Space allocated */ |
| 119 | } pipeline; |
Denys Vlasenko | 98a4c7c | 2010-02-04 15:00:15 +0100 | [diff] [blame] | 120 | } FIX_ALIASING; |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 121 | #define G (*(struct globals*)&bb_common_bufsiz1) |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 122 | struct BUG_G_too_big { |
| 123 | char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1]; |
| 124 | }; |
Denis Vlasenko | 74324c8 | 2007-06-04 10:16:52 +0000 | [diff] [blame] | 125 | #define INIT_G() do { \ |
Denis Vlasenko | 74324c8 | 2007-06-04 10:16:52 +0000 | [diff] [blame] | 126 | G.sed_cmd_tail = &G.sed_cmd_head; \ |
| 127 | } while (0) |
| 128 | |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 129 | |
Bernhard Reutner-Fischer | b7f3973 | 2006-03-01 20:14:16 +0000 | [diff] [blame] | 130 | #if ENABLE_FEATURE_CLEAN_UP |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 131 | static void sed_free_and_close_stuff(void) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 132 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 133 | sed_cmd_t *sed_cmd = G.sed_cmd_head.next; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 134 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 135 | llist_free(G.append_head, free); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 136 | |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 137 | while (sed_cmd) { |
Glenn L McGrath | bd9b32b | 2003-04-09 01:43:54 +0000 | [diff] [blame] | 138 | sed_cmd_t *sed_cmd_next = sed_cmd->next; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 139 | |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 140 | if (sed_cmd->sw_file) |
| 141 | xprint_and_close_file(sed_cmd->sw_file); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 142 | |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 143 | if (sed_cmd->beg_match) { |
| 144 | regfree(sed_cmd->beg_match); |
| 145 | free(sed_cmd->beg_match); |
Erik Andersen | e49d5ec | 2000-02-08 19:58:47 +0000 | [diff] [blame] | 146 | } |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 147 | if (sed_cmd->end_match) { |
| 148 | regfree(sed_cmd->end_match); |
| 149 | free(sed_cmd->end_match); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 150 | } |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 151 | if (sed_cmd->sub_match) { |
| 152 | regfree(sed_cmd->sub_match); |
| 153 | free(sed_cmd->sub_match); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 154 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 155 | free(sed_cmd->string); |
Glenn L McGrath | 56c633c | 2003-03-28 04:23:23 +0000 | [diff] [blame] | 156 | free(sed_cmd); |
| 157 | sed_cmd = sed_cmd_next; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 158 | } |
Rob Landley | ce4f0e9 | 2004-10-30 06:54:19 +0000 | [diff] [blame] | 159 | |
Denis Vlasenko | 6081868 | 2007-09-28 22:07:23 +0000 | [diff] [blame] | 160 | free(G.hold_space); |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 161 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 162 | while (G.current_input_file < G.input_file_count) |
| 163 | fclose(G.input_file_list[G.current_input_file++]); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 164 | } |
Denis Vlasenko | 666da5e | 2006-12-26 18:17:42 +0000 | [diff] [blame] | 165 | #else |
| 166 | void sed_free_and_close_stuff(void); |
Mark Whitley | c41e8c8 | 2000-07-12 23:35:21 +0000 | [diff] [blame] | 167 | #endif |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 168 | |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 169 | /* If something bad happens during -i operation, delete temp file */ |
| 170 | |
| 171 | static void cleanup_outname(void) |
| 172 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 173 | if (G.outname) unlink(G.outname); |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 174 | } |
| 175 | |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 176 | /* strcpy, replacing "\from" with 'to'. If to is NUL, replacing "\any" with 'any' */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 177 | |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 178 | static void parse_escapes(char *dest, const char *string, int len, char from, char to) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 179 | { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 180 | int i = 0; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 181 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 182 | while (i < len) { |
| 183 | if (string[i] == '\\') { |
| 184 | if (!to || string[i+1] == from) { |
Denis Vlasenko | 4ebaf10 | 2007-01-19 21:33:19 +0000 | [diff] [blame] | 185 | *dest++ = to ? to : string[i+1]; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 186 | i += 2; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 187 | continue; |
Denis Vlasenko | 4ebaf10 | 2007-01-19 21:33:19 +0000 | [diff] [blame] | 188 | } |
| 189 | *dest++ = string[i++]; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 190 | } |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 191 | /* TODO: is it safe wrt a string with trailing '\\' ? */ |
Denis Vlasenko | 4ebaf10 | 2007-01-19 21:33:19 +0000 | [diff] [blame] | 192 | *dest++ = string[i++]; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 193 | } |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 194 | *dest = '\0'; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 195 | } |
| 196 | |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 197 | static char *copy_parsing_escapes(const char *string, int len) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 198 | { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 199 | char *dest = xmalloc(len + 1); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 200 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 201 | parse_escapes(dest, string, len, 'n', '\n'); |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 202 | /* GNU sed also recognizes \t */ |
| 203 | parse_escapes(dest, dest, strlen(dest), 't', '\t'); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 204 | return dest; |
| 205 | } |
| 206 | |
| 207 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 208 | /* |
Eric Andersen | 28b3c53 | 2001-01-02 11:01:31 +0000 | [diff] [blame] | 209 | * index_of_next_unescaped_regexp_delim - walks left to right through a string |
| 210 | * beginning at a specified index and returns the index of the next regular |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 211 | * expression delimiter (typically a forward slash ('/')) not preceded by |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 212 | * a backslash ('\'). A negative delimiter disables square bracket checking. |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 213 | */ |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 214 | static int index_of_next_unescaped_regexp_delim(int delimiter, const char *str) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 215 | { |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 216 | int bracket = -1; |
| 217 | int escaped = 0; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 218 | int idx = 0; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 219 | char ch; |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 220 | |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 221 | if (delimiter < 0) { |
| 222 | bracket--; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 223 | delimiter = -delimiter; |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 224 | } |
| 225 | |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 226 | for (; (ch = str[idx]); idx++) { |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 227 | if (bracket >= 0) { |
Glenn L McGrath | d4185b0 | 2003-04-11 17:10:23 +0000 | [diff] [blame] | 228 | if (ch == ']' && !(bracket == idx - 1 || (bracket == idx - 2 |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 229 | && str[idx - 1] == '^'))) |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 230 | bracket = -1; |
| 231 | } else if (escaped) |
| 232 | escaped = 0; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 233 | else if (ch == '\\') |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 234 | escaped = 1; |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 235 | else if (bracket == -1 && ch == '[') |
Matt Kraai | a0065d5 | 2001-08-24 14:45:50 +0000 | [diff] [blame] | 236 | bracket = idx; |
Glenn L McGrath | 1fb4467 | 2003-03-09 08:44:49 +0000 | [diff] [blame] | 237 | else if (ch == delimiter) |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 238 | return idx; |
| 239 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 240 | |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 241 | /* if we make it to here, we've hit the end of the string */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 242 | bb_error_msg_and_die("unmatched '%c'", delimiter); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 243 | } |
| 244 | |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 245 | /* |
| 246 | * Returns the index of the third delimiter |
| 247 | */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 248 | static int parse_regex_delim(const char *cmdstr, char **match, char **replace) |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 249 | { |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 250 | const char *cmdstr_ptr = cmdstr; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 251 | char delimiter; |
| 252 | int idx = 0; |
| 253 | |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 254 | /* verify that the 's' or 'y' is followed by something. That something |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 255 | * (typically a 'slash') is now our regexp delimiter... */ |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 256 | if (*cmdstr == '\0') |
| 257 | bb_error_msg_and_die("bad format in substitution expression"); |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 258 | delimiter = *cmdstr_ptr++; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 259 | |
| 260 | /* save the match string */ |
| 261 | idx = index_of_next_unescaped_regexp_delim(delimiter, cmdstr_ptr); |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 262 | *match = copy_parsing_escapes(cmdstr_ptr, idx); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 263 | |
| 264 | /* save the replacement string */ |
| 265 | cmdstr_ptr += idx + 1; |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 266 | idx = index_of_next_unescaped_regexp_delim(-delimiter, cmdstr_ptr); |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 267 | *replace = copy_parsing_escapes(cmdstr_ptr, idx); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 268 | |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 269 | return ((cmdstr_ptr - cmdstr) + idx); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 270 | } |
| 271 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 272 | /* |
| 273 | * returns the index in the string just past where the address ends. |
| 274 | */ |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 275 | static int get_address(const char *my_str, int *linenum, regex_t ** regex) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 276 | { |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 277 | const char *pos = my_str; |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 278 | |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 279 | if (isdigit(*my_str)) { |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 280 | *linenum = strtol(my_str, (char**)&pos, 10); |
Glenn L McGrath | 1fb4467 | 2003-03-09 08:44:49 +0000 | [diff] [blame] | 281 | /* endstr shouldnt ever equal NULL */ |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 282 | } else if (*my_str == '$') { |
Mark Whitley | 0915c4b | 2001-06-11 23:50:06 +0000 | [diff] [blame] | 283 | *linenum = -1; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 284 | pos++; |
| 285 | } else if (*my_str == '/' || *my_str == '\\') { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 286 | int next; |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 287 | char delimiter; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 288 | char *temp; |
Glenn L McGrath | 1fb4467 | 2003-03-09 08:44:49 +0000 | [diff] [blame] | 289 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 290 | delimiter = '/'; |
| 291 | if (*my_str == '\\') delimiter = *++pos; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 292 | next = index_of_next_unescaped_regexp_delim(delimiter, ++pos); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 293 | temp = copy_parsing_escapes(pos, next); |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 294 | *regex = xmalloc(sizeof(regex_t)); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 295 | xregcomp(*regex, temp, G.regex_type|REG_NEWLINE); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 296 | free(temp); |
| 297 | /* Move position to next character after last delimiter */ |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 298 | pos += (next+1); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 299 | } |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 300 | return pos - my_str; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 301 | } |
Erik Andersen | e49d5ec | 2000-02-08 19:58:47 +0000 | [diff] [blame] | 302 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 303 | /* Grab a filename. Whitespace at start is skipped, then goes to EOL. */ |
Denis Vlasenko | 68404f1 | 2008-03-17 09:00:54 +0000 | [diff] [blame] | 304 | static int parse_file_cmd(/*sed_cmd_t *sed_cmd,*/ const char *filecmdstr, char **retval) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 305 | { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 306 | int start = 0, idx, hack = 0; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 307 | |
| 308 | /* Skip whitespace, then grab filename to end of line */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 309 | while (isspace(filecmdstr[start])) |
| 310 | start++; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 311 | idx = start; |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 312 | while (filecmdstr[idx] && filecmdstr[idx] != '\n') |
| 313 | idx++; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 314 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 315 | /* If lines glued together, put backslash back. */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 316 | if (filecmdstr[idx] == '\n') |
| 317 | hack = 1; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 318 | if (idx == start) |
| 319 | bb_error_msg_and_die("empty filename"); |
Rob Landley | d921b2e | 2006-08-03 15:41:12 +0000 | [diff] [blame] | 320 | *retval = xstrndup(filecmdstr+start, idx-start+hack+1); |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 321 | if (hack) |
| 322 | (*retval)[idx] = '\\'; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 323 | |
| 324 | return idx; |
| 325 | } |
| 326 | |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 327 | static int parse_subst_cmd(sed_cmd_t *sed_cmd, const char *substr) |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 328 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 329 | int cflags = G.regex_type; |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 330 | char *match; |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 331 | int idx; |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 332 | |
| 333 | /* |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 334 | * A substitution command should look something like this: |
| 335 | * s/match/replace/ #gIpw |
Mark Whitley | 0e4cec0 | 2000-08-21 21:29:20 +0000 | [diff] [blame] | 336 | * || | ||| |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 337 | * mandatory optional |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 338 | */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 339 | idx = parse_regex_delim(substr, &match, &sed_cmd->string); |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 340 | |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 341 | /* determine the number of back references in the match string */ |
| 342 | /* Note: we compute this here rather than in the do_subst_command() |
| 343 | * function to save processor time, at the expense of a little more memory |
| 344 | * (4 bits) per sed_cmd */ |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 345 | |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 346 | /* process the flags */ |
Mark Whitley | 70705d7 | 2000-07-14 19:06:30 +0000 | [diff] [blame] | 347 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 348 | sed_cmd->which_match = 1; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 349 | while (substr[++idx]) { |
| 350 | /* Parse match number */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 351 | if (isdigit(substr[idx])) { |
| 352 | if (match[0] != '^') { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 353 | /* Match 0 treated as all, multiple matches we take the last one. */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 354 | const char *pos = substr + idx; |
| 355 | /* FIXME: error check? */ |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 356 | sed_cmd->which_match = (unsigned)strtol(substr+idx, (char**) &pos, 10); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 357 | idx = pos - substr; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 358 | } |
| 359 | continue; |
| 360 | } |
Rob Landley | 40ec4ae | 2004-01-04 06:42:14 +0000 | [diff] [blame] | 361 | /* Skip spaces */ |
Denys Vlasenko | 6935ec9 | 2009-10-22 19:42:26 +0200 | [diff] [blame] | 362 | if (isspace(substr[idx])) |
| 363 | continue; |
Rob Landley | 40ec4ae | 2004-01-04 06:42:14 +0000 | [diff] [blame] | 364 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 365 | switch (substr[idx]) { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 366 | /* Replace all occurrences */ |
| 367 | case 'g': |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 368 | if (match[0] != '^') |
| 369 | sed_cmd->which_match = 0; |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 370 | break; |
| 371 | /* Print pattern space */ |
| 372 | case 'p': |
| 373 | sed_cmd->sub_p = 1; |
| 374 | break; |
| 375 | /* Write to file */ |
| 376 | case 'w': |
| 377 | { |
| 378 | char *temp; |
Denis Vlasenko | 68404f1 | 2008-03-17 09:00:54 +0000 | [diff] [blame] | 379 | idx += parse_file_cmd(/*sed_cmd,*/ substr+idx, &temp); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 380 | break; |
| 381 | } |
| 382 | /* Ignore case (gnu exension) */ |
| 383 | case 'I': |
| 384 | cflags |= REG_ICASE; |
| 385 | break; |
| 386 | /* Comment */ |
| 387 | case '#': |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 388 | // while (substr[++idx]) continue; |
| 389 | idx += strlen(substr + idx); // same |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 390 | /* Fall through */ |
| 391 | /* End of command */ |
| 392 | case ';': |
| 393 | case '}': |
| 394 | goto out; |
| 395 | default: |
| 396 | bb_error_msg_and_die("bad option in substitution expression"); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 397 | } |
| 398 | } |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 399 | out: |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 400 | /* compile the match string into a regex */ |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 401 | if (*match != '\0') { |
| 402 | /* If match is empty, we use last regex used at runtime */ |
Denis Vlasenko | b95636c | 2006-12-19 23:36:04 +0000 | [diff] [blame] | 403 | sed_cmd->sub_match = xmalloc(sizeof(regex_t)); |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 404 | xregcomp(sed_cmd->sub_match, match, cflags); |
| 405 | } |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 406 | free(match); |
Mark Whitley | 70705d7 | 2000-07-14 19:06:30 +0000 | [diff] [blame] | 407 | |
| 408 | return idx; |
Mark Whitley | 06f3529 | 2000-07-13 19:58:04 +0000 | [diff] [blame] | 409 | } |
| 410 | |
Glenn L McGrath | 2971ef1 | 2003-03-18 01:19:23 +0000 | [diff] [blame] | 411 | /* |
| 412 | * Process the commands arguments |
| 413 | */ |
Denis Vlasenko | 9356b50 | 2007-01-29 23:44:38 +0000 | [diff] [blame] | 414 | static const char *parse_cmd_args(sed_cmd_t *sed_cmd, const char *cmdstr) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 415 | { |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 416 | /* handle (s)ubstitution command */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 417 | if (sed_cmd->cmd == 's') |
| 418 | cmdstr += parse_subst_cmd(sed_cmd, cmdstr); |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 419 | /* handle edit cmds: (a)ppend, (i)nsert, and (c)hange */ |
| 420 | else if (strchr("aic", sed_cmd->cmd)) { |
| 421 | if ((sed_cmd->end_line || sed_cmd->end_match) && sed_cmd->cmd != 'c') |
Denys Vlasenko | 6935ec9 | 2009-10-22 19:42:26 +0200 | [diff] [blame] | 422 | bb_error_msg_and_die("only a beginning address can be specified for edit commands"); |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 423 | for (;;) { |
| 424 | if (*cmdstr == '\n' || *cmdstr == '\\') { |
Rob Landley | 25d8239 | 2004-04-01 09:23:30 +0000 | [diff] [blame] | 425 | cmdstr++; |
| 426 | break; |
Denys Vlasenko | c0dab37 | 2009-10-22 22:28:08 +0200 | [diff] [blame] | 427 | } |
| 428 | if (!isspace(*cmdstr)) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 429 | break; |
Denys Vlasenko | c0dab37 | 2009-10-22 22:28:08 +0200 | [diff] [blame] | 430 | cmdstr++; |
Rob Landley | 25d8239 | 2004-04-01 09:23:30 +0000 | [diff] [blame] | 431 | } |
Rob Landley | d921b2e | 2006-08-03 15:41:12 +0000 | [diff] [blame] | 432 | sed_cmd->string = xstrdup(cmdstr); |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 433 | /* "\anychar" -> "anychar" */ |
| 434 | parse_escapes(sed_cmd->string, sed_cmd->string, strlen(cmdstr), '\0', '\0'); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 435 | cmdstr += strlen(cmdstr); |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 436 | /* handle file cmds: (r)ead */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 437 | } else if (strchr("rw", sed_cmd->cmd)) { |
Glenn L McGrath | 2f8a401 | 2003-03-09 02:44:49 +0000 | [diff] [blame] | 438 | if (sed_cmd->end_line || sed_cmd->end_match) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 439 | bb_error_msg_and_die("command only uses one address"); |
Denis Vlasenko | 68404f1 | 2008-03-17 09:00:54 +0000 | [diff] [blame] | 440 | cmdstr += parse_file_cmd(/*sed_cmd,*/ cmdstr, &sed_cmd->string); |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 441 | if (sed_cmd->cmd == 'w') { |
Denis Vlasenko | 5415c85 | 2008-07-21 23:05:26 +0000 | [diff] [blame] | 442 | sed_cmd->sw_file = xfopen_for_write(sed_cmd->string); |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 443 | sed_cmd->sw_last_char = '\n'; |
| 444 | } |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 445 | /* handle branch commands */ |
Rob Landley | 93850a5 | 2005-05-18 06:34:37 +0000 | [diff] [blame] | 446 | } else if (strchr(":btT", sed_cmd->cmd)) { |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 447 | int length; |
| 448 | |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 449 | cmdstr = skip_whitespace(cmdstr); |
Glenn L McGrath | f452356 | 2003-09-14 06:01:14 +0000 | [diff] [blame] | 450 | length = strcspn(cmdstr, semicolon_whitespace); |
| 451 | if (length) { |
Rob Landley | d921b2e | 2006-08-03 15:41:12 +0000 | [diff] [blame] | 452 | sed_cmd->string = xstrndup(cmdstr, length); |
Glenn L McGrath | f452356 | 2003-09-14 06:01:14 +0000 | [diff] [blame] | 453 | cmdstr += length; |
| 454 | } |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 455 | } |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 456 | /* translation command */ |
| 457 | else if (sed_cmd->cmd == 'y') { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 458 | char *match, *replace; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 459 | int i = cmdstr[0]; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 460 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 461 | cmdstr += parse_regex_delim(cmdstr, &match, &replace)+1; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 462 | /* \n already parsed, but \delimiter needs unescaping. */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 463 | parse_escapes(match, match, strlen(match), i, i); |
| 464 | parse_escapes(replace, replace, strlen(replace), i, i); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 465 | |
Rob Landley | 9ffd423 | 2006-05-21 18:30:35 +0000 | [diff] [blame] | 466 | sed_cmd->string = xzalloc((strlen(match) + 1) * 2); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 467 | for (i = 0; match[i] && replace[i]; i++) { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 468 | sed_cmd->string[i*2] = match[i]; |
| 469 | sed_cmd->string[i*2+1] = replace[i]; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 470 | } |
| 471 | free(match); |
| 472 | free(replace); |
Glenn L McGrath | f01b46d | 2003-03-30 08:02:18 +0000 | [diff] [blame] | 473 | } |
Glenn L McGrath | 2971ef1 | 2003-03-18 01:19:23 +0000 | [diff] [blame] | 474 | /* if it wasnt a single-letter command that takes no arguments |
| 475 | * then it must be an invalid command. |
| 476 | */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 477 | else if (strchr("dDgGhHlnNpPqx={}", sed_cmd->cmd) == 0) { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 478 | bb_error_msg_and_die("unsupported command %c", sed_cmd->cmd); |
Mark Whitley | 1f3b9f2 | 2001-05-11 22:27:13 +0000 | [diff] [blame] | 479 | } |
Mark Whitley | 70705d7 | 2000-07-14 19:06:30 +0000 | [diff] [blame] | 480 | |
| 481 | /* give back whatever's left over */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 482 | return cmdstr; |
Eric Andersen | 50d6360 | 1999-11-09 01:47:36 +0000 | [diff] [blame] | 483 | } |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 484 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 485 | |
| 486 | /* Parse address+command sets, skipping comment lines. */ |
| 487 | |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 488 | static void add_cmd(const char *cmdstr) |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 489 | { |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 490 | sed_cmd_t *sed_cmd; |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 491 | unsigned len, n; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 492 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 493 | /* Append this line to any unfinished line from last time. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 494 | if (G.add_cmd_line) { |
| 495 | char *tp = xasprintf("%s\n%s", G.add_cmd_line, cmdstr); |
| 496 | free(G.add_cmd_line); |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 497 | cmdstr = G.add_cmd_line = tp; |
Rob Landley | 12d8755 | 2006-06-05 17:32:44 +0000 | [diff] [blame] | 498 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 499 | |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 500 | /* If this line ends with unescaped backslash, request next line. */ |
| 501 | n = len = strlen(cmdstr); |
| 502 | while (n && cmdstr[n-1] == '\\') |
| 503 | n--; |
| 504 | if ((len - n) & 1) { /* if odd number of trailing backslashes */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 505 | if (!G.add_cmd_line) |
| 506 | G.add_cmd_line = xstrdup(cmdstr); |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 507 | G.add_cmd_line[len-1] = '\0'; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 508 | return; |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 509 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 510 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 511 | /* Loop parsing all commands in this line. */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 512 | while (*cmdstr) { |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 513 | /* Skip leading whitespace and semicolons */ |
| 514 | cmdstr += strspn(cmdstr, semicolon_whitespace); |
| 515 | |
| 516 | /* If no more commands, exit. */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 517 | if (!*cmdstr) break; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 518 | |
| 519 | /* if this is a comment, jump past it and keep going */ |
| 520 | if (*cmdstr == '#') { |
| 521 | /* "#n" is the same as using -n on the command line */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 522 | if (cmdstr[1] == 'n') |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 523 | G.be_quiet++; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 524 | cmdstr = strpbrk(cmdstr, "\n\r"); |
| 525 | if (!cmdstr) break; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 526 | continue; |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 527 | } |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 528 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 529 | /* parse the command |
| 530 | * format is: [addr][,addr][!]cmd |
| 531 | * |----||-----||-| |
| 532 | * part1 part2 part3 |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 533 | */ |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 534 | |
Rob Landley | 9ffd423 | 2006-05-21 18:30:35 +0000 | [diff] [blame] | 535 | sed_cmd = xzalloc(sizeof(sed_cmd_t)); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 536 | |
| 537 | /* first part (if present) is an address: either a '$', a number or a /regex/ */ |
| 538 | cmdstr += get_address(cmdstr, &sed_cmd->beg_line, &sed_cmd->beg_match); |
| 539 | |
| 540 | /* second part (if present) will begin with a comma */ |
| 541 | if (*cmdstr == ',') { |
| 542 | int idx; |
| 543 | |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 544 | cmdstr++; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 545 | idx = get_address(cmdstr, &sed_cmd->end_line, &sed_cmd->end_match); |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 546 | if (!idx) |
| 547 | bb_error_msg_and_die("no address after comma"); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 548 | cmdstr += idx; |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 549 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 550 | |
| 551 | /* skip whitespace before the command */ |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 552 | cmdstr = skip_whitespace(cmdstr); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 553 | |
| 554 | /* Check for inversion flag */ |
| 555 | if (*cmdstr == '!') { |
| 556 | sed_cmd->invert = 1; |
| 557 | cmdstr++; |
| 558 | |
| 559 | /* skip whitespace before the command */ |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 560 | cmdstr = skip_whitespace(cmdstr); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 561 | } |
| 562 | |
| 563 | /* last part (mandatory) will be a command */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 564 | if (!*cmdstr) |
| 565 | bb_error_msg_and_die("missing command"); |
Denys Vlasenko | a2215b9 | 2010-05-12 01:49:04 +0200 | [diff] [blame] | 566 | sed_cmd->cmd = *cmdstr++; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 567 | cmdstr = parse_cmd_args(sed_cmd, cmdstr); |
| 568 | |
| 569 | /* Add the command to the command array */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 570 | G.sed_cmd_tail->next = sed_cmd; |
| 571 | G.sed_cmd_tail = G.sed_cmd_tail->next; |
Glenn L McGrath | f50ce31 | 2003-03-09 15:12:24 +0000 | [diff] [blame] | 572 | } |
| 573 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 574 | /* If we glued multiple lines together, free the memory. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 575 | free(G.add_cmd_line); |
| 576 | G.add_cmd_line = NULL; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 577 | } |
| 578 | |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 579 | /* Append to a string, reallocating memory as necessary. */ |
| 580 | |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 581 | #define PIPE_GROW 64 |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 582 | |
Eric Andersen | 14f5c8d | 2005-04-16 19:39:00 +0000 | [diff] [blame] | 583 | static void pipe_putc(char c) |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 584 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 585 | if (G.pipeline.idx == G.pipeline.len) { |
| 586 | G.pipeline.buf = xrealloc(G.pipeline.buf, |
| 587 | G.pipeline.len + PIPE_GROW); |
| 588 | G.pipeline.len += PIPE_GROW; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 589 | } |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 590 | G.pipeline.buf[G.pipeline.idx++] = c; |
Eric Andersen | c52a6b0 | 2001-11-10 10:49:42 +0000 | [diff] [blame] | 591 | } |
| 592 | |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 593 | static void do_subst_w_backrefs(char *line, char *replace) |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 594 | { |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 595 | int i, j; |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 596 | |
| 597 | /* go through the replacement string */ |
| 598 | for (i = 0; replace[i]; i++) { |
| 599 | /* if we find a backreference (\1, \2, etc.) print the backref'ed * text */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 600 | if (replace[i] == '\\') { |
| 601 | unsigned backref = replace[++i] - '0'; |
| 602 | if (backref <= 9) { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 603 | /* print out the text held in G.regmatch[backref] */ |
| 604 | if (G.regmatch[backref].rm_so != -1) { |
| 605 | j = G.regmatch[backref].rm_so; |
| 606 | while (j < G.regmatch[backref].rm_eo) |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 607 | pipe_putc(line[j++]); |
| 608 | } |
| 609 | continue; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 610 | } |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 611 | /* I _think_ it is impossible to get '\' to be |
| 612 | * the last char in replace string. Thus we dont check |
| 613 | * for replace[i] == NUL. (counterexample anyone?) */ |
| 614 | /* if we find a backslash escaped character, print the character */ |
| 615 | pipe_putc(replace[i]); |
| 616 | continue; |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 617 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 618 | /* if we find an unescaped '&' print out the whole matched text. */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 619 | if (replace[i] == '&') { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 620 | j = G.regmatch[0].rm_so; |
| 621 | while (j < G.regmatch[0].rm_eo) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 622 | pipe_putc(line[j++]); |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 623 | continue; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 624 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 625 | /* Otherwise just output the character. */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 626 | pipe_putc(replace[i]); |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 627 | } |
| 628 | } |
| 629 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 630 | static int do_subst_command(sed_cmd_t *sed_cmd, char **line_p) |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 631 | { |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 632 | char *line = *line_p; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 633 | int altered = 0; |
Denis Vlasenko | 77ad97f | 2008-05-13 02:27:31 +0000 | [diff] [blame] | 634 | unsigned match_count = 0; |
Glenn L McGrath | c1d9507 | 2003-04-08 06:42:45 +0000 | [diff] [blame] | 635 | regex_t *current_regex; |
| 636 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 637 | current_regex = sed_cmd->sub_match; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 638 | /* Handle empty regex. */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 639 | if (!current_regex) { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 640 | current_regex = G.previous_regex_ptr; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 641 | if (!current_regex) |
Denis Vlasenko | d3d004d | 2006-10-27 09:02:31 +0000 | [diff] [blame] | 642 | bb_error_msg_and_die("no previous regexp"); |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 643 | } |
| 644 | G.previous_regex_ptr = current_regex; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 645 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 646 | /* Find the first match */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 647 | if (REG_NOMATCH == regexec(current_regex, line, 10, G.regmatch, 0)) |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 648 | return 0; |
| 649 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 650 | /* Initialize temporary output buffer. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 651 | G.pipeline.buf = xmalloc(PIPE_GROW); |
| 652 | G.pipeline.len = PIPE_GROW; |
| 653 | G.pipeline.idx = 0; |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 654 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 655 | /* Now loop through, substituting for matches */ |
| 656 | do { |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 657 | int i; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 658 | |
Eric Andersen | c06f568 | 2004-02-04 10:57:46 +0000 | [diff] [blame] | 659 | /* Work around bug in glibc regexec, demonstrated by: |
| 660 | echo " a.b" | busybox sed 's [^ .]* x g' |
| 661 | The match_count check is so not to break |
| 662 | echo "hi" | busybox sed 's/^/!/g' */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 663 | if (!G.regmatch[0].rm_so && !G.regmatch[0].rm_eo && match_count) { |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 664 | pipe_putc(*line++); |
Eric Andersen | c06f568 | 2004-02-04 10:57:46 +0000 | [diff] [blame] | 665 | continue; |
| 666 | } |
| 667 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 668 | match_count++; |
| 669 | |
| 670 | /* If we aren't interested in this match, output old line to |
| 671 | end of match and continue */ |
Denis Vlasenko | 77ad97f | 2008-05-13 02:27:31 +0000 | [diff] [blame] | 672 | if (sed_cmd->which_match |
| 673 | && (sed_cmd->which_match != match_count) |
| 674 | ) { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 675 | for (i = 0; i < G.regmatch[0].rm_eo; i++) |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 676 | pipe_putc(*line++); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 677 | continue; |
| 678 | } |
| 679 | |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 680 | /* print everything before the match */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 681 | for (i = 0; i < G.regmatch[0].rm_so; i++) |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 682 | pipe_putc(line[i]); |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 683 | |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 684 | /* then print the substitution string */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 685 | do_subst_w_backrefs(line, sed_cmd->string); |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 686 | |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 687 | /* advance past the match */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 688 | line += G.regmatch[0].rm_eo; |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 689 | /* flag that something has changed */ |
| 690 | altered++; |
Mark Whitley | 97562bd | 2000-07-17 20:06:42 +0000 | [diff] [blame] | 691 | |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 692 | /* if we're not doing this globally, get out now */ |
Denis Vlasenko | 284d0fa | 2008-02-16 13:18:17 +0000 | [diff] [blame] | 693 | if (sed_cmd->which_match) |
| 694 | break; |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 695 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 696 | //maybe (G.regmatch[0].rm_eo ? REG_NOTBOL : 0) instead of unconditional REG_NOTBOL? |
Denys Vlasenko | 11c8238 | 2009-09-22 03:02:21 +0200 | [diff] [blame] | 697 | } while (*line && regexec(current_regex, line, 10, G.regmatch, REG_NOTBOL) != REG_NOMATCH); |
Mark Whitley | 2dc192f | 2000-11-03 19:47:00 +0000 | [diff] [blame] | 698 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 699 | /* Copy rest of string into output pipeline */ |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 700 | while (1) { |
| 701 | char c = *line++; |
| 702 | pipe_putc(c); |
| 703 | if (c == '\0') |
| 704 | break; |
| 705 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 706 | |
Denys Vlasenko | f210cff | 2009-08-17 01:35:04 +0200 | [diff] [blame] | 707 | free(*line_p); |
| 708 | *line_p = G.pipeline.buf; |
Mark Whitley | 4f7fe77 | 2000-07-13 20:01:58 +0000 | [diff] [blame] | 709 | return altered; |
| 710 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 711 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 712 | /* Set command pointer to point to this label. (Does not handle null label.) */ |
Rob Landley | 4795e4e | 2006-07-26 17:25:08 +0000 | [diff] [blame] | 713 | static sed_cmd_t *branch_to(char *label) |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 714 | { |
| 715 | sed_cmd_t *sed_cmd; |
Glenn L McGrath | d4185b0 | 2003-04-11 17:10:23 +0000 | [diff] [blame] | 716 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 717 | for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 718 | if (sed_cmd->cmd == ':' && sed_cmd->string && !strcmp(sed_cmd->string, label)) { |
| 719 | return sed_cmd; |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 720 | } |
| 721 | } |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 722 | bb_error_msg_and_die("can't find label for jump to '%s'", label); |
Glenn L McGrath | 961c6c1 | 2003-03-28 04:43:39 +0000 | [diff] [blame] | 723 | } |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 724 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 725 | static void append(char *s) |
| 726 | { |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 727 | llist_add_to_end(&G.append_head, xstrdup(s)); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 728 | } |
| 729 | |
| 730 | static void flush_append(void) |
| 731 | { |
Rob Landley | 0b65628 | 2006-05-08 22:17:23 +0000 | [diff] [blame] | 732 | char *data; |
| 733 | |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 734 | /* Output appended lines. */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 735 | while ((data = (char *)llist_pop(&G.append_head))) { |
| 736 | fprintf(G.nonstdout, "%s\n", data); |
Rob Landley | 0b65628 | 2006-05-08 22:17:23 +0000 | [diff] [blame] | 737 | free(data); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 738 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 739 | } |
| 740 | |
Eric Andersen | 14f5c8d | 2005-04-16 19:39:00 +0000 | [diff] [blame] | 741 | static void add_input_file(FILE *file) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 742 | { |
Denis Vlasenko | deeed59 | 2008-07-08 05:14:36 +0000 | [diff] [blame] | 743 | G.input_file_list = xrealloc_vector(G.input_file_list, 2, G.input_file_count); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 744 | G.input_file_list[G.input_file_count++] = file; |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 745 | } |
| 746 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 747 | /* Get next line of input from G.input_file_list, flushing append buffer and |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 748 | * noting if we ran out of files without a newline on the last line we read. |
| 749 | */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 750 | enum { |
| 751 | NO_EOL_CHAR = 1, |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 752 | LAST_IS_NUL = 2, |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 753 | }; |
| 754 | static char *get_next_line(char *gets_char) |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 755 | { |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 756 | char *temp = NULL; |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 757 | int len; |
| 758 | char gc; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 759 | |
| 760 | flush_append(); |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 761 | |
| 762 | /* will be returned if last line in the file |
| 763 | * doesn't end with either '\n' or '\0' */ |
| 764 | gc = NO_EOL_CHAR; |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 765 | while (G.current_input_file < G.input_file_count) { |
| 766 | FILE *fp = G.input_file_list[G.current_input_file]; |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 767 | /* Read line up to a newline or NUL byte, inclusive, |
| 768 | * return malloc'ed char[]. length of the chunk read |
| 769 | * is stored in len. NULL if EOF/error */ |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 770 | temp = bb_get_chunk_from_file(fp, &len); |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 771 | if (temp) { |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 772 | /* len > 0 here, it's ok to do temp[len-1] */ |
| 773 | char c = temp[len-1]; |
| 774 | if (c == '\n' || c == '\0') { |
| 775 | temp[len-1] = '\0'; |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 776 | gc = c; |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 777 | if (c == '\0') { |
| 778 | int ch = fgetc(fp); |
| 779 | if (ch != EOF) |
| 780 | ungetc(ch, fp); |
| 781 | else |
| 782 | gc = LAST_IS_NUL; |
| 783 | } |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 784 | } |
Denis Vlasenko | c562bb7 | 2007-01-29 17:08:51 +0000 | [diff] [blame] | 785 | /* else we put NO_EOL_CHAR into *gets_char */ |
| 786 | break; |
| 787 | |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 788 | /* NB: I had the idea of peeking next file(s) and returning |
| 789 | * NO_EOL_CHAR only if it is the *last* non-empty |
| 790 | * input file. But there is a case where this won't work: |
| 791 | * file1: "a woo\nb woo" |
| 792 | * file2: "c no\nd no" |
| 793 | * sed -ne 's/woo/bang/p' input1 input2 => "a bang\nb bang" |
| 794 | * (note: *no* newline after "b bang"!) */ |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 795 | } |
| 796 | /* Close this file and advance to next one */ |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 797 | fclose(fp); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 798 | G.current_input_file++; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 799 | } |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 800 | *gets_char = gc; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 801 | return temp; |
| 802 | } |
| 803 | |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 804 | /* Output line of text. */ |
| 805 | /* Note: |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 806 | * The tricks with NO_EOL_CHAR and last_puts_char are there to emulate gnu sed. |
Denis Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 807 | * Without them, we had this: |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 808 | * echo -n thingy >z1 |
| 809 | * echo -n again >z2 |
| 810 | * >znull |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 811 | * sed "s/i/z/" z1 z2 znull | hexdump -vC |
| 812 | * output: |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 813 | * gnu sed 4.1.5: |
| 814 | * 00000000 74 68 7a 6e 67 79 0a 61 67 61 7a 6e |thzngy.agazn| |
| 815 | * bbox: |
| 816 | * 00000000 74 68 7a 6e 67 79 61 67 61 7a 6e |thzngyagazn| |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 817 | */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 818 | static void puts_maybe_newline(char *s, FILE *file, char *last_puts_char, char last_gets_char) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 819 | { |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 820 | char lpc = *last_puts_char; |
Denis Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 821 | |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 822 | /* Need to insert a '\n' between two files because first file's |
| 823 | * last line wasn't terminated? */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 824 | if (lpc != '\n' && lpc != '\0') { |
Denis Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 825 | fputc('\n', file); |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 826 | lpc = '\n'; |
Denis Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 827 | } |
Denis Vlasenko | 8b22b07 | 2006-12-02 17:58:10 +0000 | [diff] [blame] | 828 | fputs(s, file); |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 829 | |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 830 | /* 'x' - just something which is not '\n', '\0' or NO_EOL_CHAR */ |
Denis Vlasenko | ef44d9d | 2007-01-17 23:16:16 +0000 | [diff] [blame] | 831 | if (s[0]) |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 832 | lpc = 'x'; |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 833 | |
| 834 | /* had trailing '\0' and it was last char of file? */ |
| 835 | if (last_gets_char == LAST_IS_NUL) { |
| 836 | fputc('\0', file); |
| 837 | lpc = 'x'; /* */ |
| 838 | } else |
| 839 | /* had trailing '\n' or '\0'? */ |
| 840 | if (last_gets_char != NO_EOL_CHAR) { |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 841 | fputc(last_gets_char, file); |
| 842 | lpc = last_gets_char; |
Denis Vlasenko | 1375bc7 | 2006-12-02 20:12:12 +0000 | [diff] [blame] | 843 | } |
Denis Vlasenko | 8681180 | 2007-01-29 17:10:19 +0000 | [diff] [blame] | 844 | |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 845 | if (ferror(file)) { |
Denis Vlasenko | 4092082 | 2006-10-03 20:28:06 +0000 | [diff] [blame] | 846 | xfunc_error_retval = 4; /* It's what gnu sed exits with... */ |
Bernhard Reutner-Fischer | a3d4bf3 | 2006-06-03 21:40:11 +0000 | [diff] [blame] | 847 | bb_error_msg_and_die(bb_msg_write_error); |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 848 | } |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 849 | *last_puts_char = lpc; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 850 | } |
| 851 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 852 | #define sed_puts(s, n) (puts_maybe_newline(s, G.nonstdout, &last_puts_char, n)) |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 853 | |
Denis Vlasenko | 8274e06 | 2007-08-06 03:41:08 +0000 | [diff] [blame] | 854 | static int beg_match(sed_cmd_t *sed_cmd, const char *pattern_space) |
| 855 | { |
| 856 | int retval = sed_cmd->beg_match && !regexec(sed_cmd->beg_match, pattern_space, 0, NULL, 0); |
| 857 | if (retval) |
| 858 | G.previous_regex_ptr = sed_cmd->beg_match; |
| 859 | return retval; |
| 860 | } |
| 861 | |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 862 | /* Process all the lines in all the files */ |
| 863 | |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 864 | static void process_files(void) |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 865 | { |
Rob Landley | ce4f0e9 | 2004-10-30 06:54:19 +0000 | [diff] [blame] | 866 | char *pattern_space, *next_line; |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 867 | int linenum = 0; |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 868 | char last_puts_char = '\n'; |
| 869 | char last_gets_char, next_gets_char; |
Denis Vlasenko | 2ea630f | 2006-12-10 02:52:19 +0000 | [diff] [blame] | 870 | sed_cmd_t *sed_cmd; |
| 871 | int substituted; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 872 | |
Rob Landley | 2b26fd5 | 2006-02-24 02:30:39 +0000 | [diff] [blame] | 873 | /* Prime the pump */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 874 | next_line = get_next_line(&next_gets_char); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 875 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 876 | /* Go through every line in each file */ |
Denis Vlasenko | f39c7c0 | 2008-02-28 17:59:01 +0000 | [diff] [blame] | 877 | again: |
Denis Vlasenko | 2ea630f | 2006-12-10 02:52:19 +0000 | [diff] [blame] | 878 | substituted = 0; |
Glenn L McGrath | 505bd0f | 2003-03-08 05:21:02 +0000 | [diff] [blame] | 879 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 880 | /* Advance to next line. Stop if out of lines. */ |
| 881 | pattern_space = next_line; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 882 | if (!pattern_space) |
| 883 | return; |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 884 | last_gets_char = next_gets_char; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 885 | |
| 886 | /* Read one line in advance so we can act on the last line, |
| 887 | * the '$' address */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 888 | next_line = get_next_line(&next_gets_char); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 889 | linenum++; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 890 | |
| 891 | /* For every line, go through all the commands */ |
Denis Vlasenko | f39c7c0 | 2008-02-28 17:59:01 +0000 | [diff] [blame] | 892 | restart: |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 893 | for (sed_cmd = G.sed_cmd_head.next; sed_cmd; sed_cmd = sed_cmd->next) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 894 | int old_matched, matched; |
Eric Andersen | 52a3c27 | 2003-12-23 08:53:51 +0000 | [diff] [blame] | 895 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 896 | old_matched = sed_cmd->in_match; |
Mark Whitley | 0915c4b | 2001-06-11 23:50:06 +0000 | [diff] [blame] | 897 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 898 | /* Determine if this command matches this line: */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 899 | |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 900 | //bb_error_msg("match1:%d", sed_cmd->in_match); |
| 901 | //bb_error_msg("match2:%d", (!sed_cmd->beg_line && !sed_cmd->end_line |
| 902 | // && !sed_cmd->beg_match && !sed_cmd->end_match)); |
| 903 | //bb_error_msg("match3:%d", (sed_cmd->beg_line > 0 |
| 904 | // && (sed_cmd->end_line || sed_cmd->end_match |
| 905 | // ? (sed_cmd->beg_line <= linenum) |
| 906 | // : (sed_cmd->beg_line == linenum) |
| 907 | // ) |
| 908 | // ) |
| 909 | //bb_error_msg("match4:%d", (beg_match(sed_cmd, pattern_space))); |
| 910 | //bb_error_msg("match5:%d", (sed_cmd->beg_line == -1 && next_line == NULL)); |
| 911 | |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 912 | /* Are we continuing a previous multi-line match? */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 913 | sed_cmd->in_match = sed_cmd->in_match |
| 914 | /* Or is no range necessary? */ |
| 915 | || (!sed_cmd->beg_line && !sed_cmd->end_line |
| 916 | && !sed_cmd->beg_match && !sed_cmd->end_match) |
| 917 | /* Or did we match the start of a numerical range? */ |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 918 | || (sed_cmd->beg_line > 0 |
| 919 | && (sed_cmd->end_line || sed_cmd->end_match |
Denys Vlasenko | c467924 | 2010-06-04 01:31:48 +0200 | [diff] [blame] | 920 | /* note: even if end is numeric and is < linenum too, |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 921 | * GNU sed matches! We match too */ |
| 922 | ? (sed_cmd->beg_line <= linenum) /* N,end */ |
| 923 | : (sed_cmd->beg_line == linenum) /* N */ |
| 924 | ) |
| 925 | ) |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 926 | /* Or does this line match our begin address regex? */ |
Denis Vlasenko | 8274e06 | 2007-08-06 03:41:08 +0000 | [diff] [blame] | 927 | || (beg_match(sed_cmd, pattern_space)) |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 928 | /* Or did we match last line of input? */ |
| 929 | || (sed_cmd->beg_line == -1 && next_line == NULL); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 930 | |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 931 | /* Snapshot the value */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 932 | matched = sed_cmd->in_match; |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 933 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 934 | //bb_error_msg("cmd:'%c' matched:%d beg_line:%d end_line:%d linenum:%d", |
| 935 | //sed_cmd->cmd, matched, sed_cmd->beg_line, sed_cmd->end_line, linenum); |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 936 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 937 | /* Is this line the end of the current match? */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 938 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 939 | if (matched) { |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 940 | /* once matched, "n,xxx" range is dead, disabling it */ |
| 941 | if (sed_cmd->beg_line > 0) |
| 942 | sed_cmd->beg_line = -2; |
| 943 | sed_cmd->in_match = !( |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 944 | /* has the ending line come, or is this a single address command? */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 945 | (sed_cmd->end_line ? |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 946 | sed_cmd->end_line == -1 ? |
| 947 | !next_line |
| 948 | : (sed_cmd->end_line <= linenum) |
| 949 | : !sed_cmd->end_match |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 950 | ) |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 951 | /* or does this line matches our last address regex */ |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 952 | || (sed_cmd->end_match && old_matched |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 953 | && (regexec(sed_cmd->end_match, |
Denys Vlasenko | 8e96b5b | 2009-07-14 01:02:57 +0200 | [diff] [blame] | 954 | pattern_space, 0, NULL, 0) == 0)) |
| 955 | ); |
Erik Andersen | e49d5ec | 2000-02-08 19:58:47 +0000 | [diff] [blame] | 956 | } |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 957 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 958 | /* Skip blocks of commands we didn't match */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 959 | if (sed_cmd->cmd == '{') { |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 960 | if (sed_cmd->invert ? matched : !matched) { |
Denys Vlasenko | f2c16ed | 2010-04-20 04:00:03 -0400 | [diff] [blame] | 961 | unsigned nest_cnt = 0; |
| 962 | while (1) { |
| 963 | if (sed_cmd->cmd == '{') |
| 964 | nest_cnt++; |
| 965 | if (sed_cmd->cmd == '}') { |
| 966 | nest_cnt--; |
| 967 | if (nest_cnt == 0) |
| 968 | break; |
| 969 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 970 | sed_cmd = sed_cmd->next; |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 971 | if (!sed_cmd) |
| 972 | bb_error_msg_and_die("unterminated {"); |
| 973 | } |
| 974 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 975 | continue; |
| 976 | } |
Glenn L McGrath | aa5a602 | 2003-10-01 03:06:16 +0000 | [diff] [blame] | 977 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 978 | /* Okay, so did this line match? */ |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 979 | if (sed_cmd->invert ? matched : !matched) |
| 980 | continue; /* no */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 981 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 982 | /* Update last used regex in case a blank substitute BRE is found */ |
| 983 | if (sed_cmd->beg_match) { |
| 984 | G.previous_regex_ptr = sed_cmd->beg_match; |
| 985 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 986 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 987 | /* actual sedding */ |
| 988 | switch (sed_cmd->cmd) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 989 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 990 | /* Print line number */ |
| 991 | case '=': |
| 992 | fprintf(G.nonstdout, "%d\n", linenum); |
| 993 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 994 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 995 | /* Write the current pattern space up to the first newline */ |
| 996 | case 'P': |
| 997 | { |
| 998 | char *tmp = strchr(pattern_space, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 999 | if (tmp) { |
| 1000 | *tmp = '\0'; |
| 1001 | /* TODO: explain why '\n' below */ |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1002 | sed_puts(pattern_space, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1003 | *tmp = '\n'; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1004 | break; |
| 1005 | } |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1006 | /* Fall Through */ |
| 1007 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1008 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1009 | /* Write the current pattern space to output */ |
| 1010 | case 'p': |
| 1011 | /* NB: we print this _before_ the last line |
| 1012 | * (of current file) is printed. Even if |
| 1013 | * that line is nonterminated, we print |
| 1014 | * '\n' here (gnu sed does the same) */ |
| 1015 | sed_puts(pattern_space, '\n'); |
| 1016 | break; |
| 1017 | /* Delete up through first newline */ |
| 1018 | case 'D': |
| 1019 | { |
| 1020 | char *tmp = strchr(pattern_space, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1021 | if (tmp) { |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1022 | overlapping_strcpy(pattern_space, tmp + 1); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1023 | goto restart; |
| 1024 | } |
| 1025 | } |
| 1026 | /* discard this line. */ |
| 1027 | case 'd': |
| 1028 | goto discard_line; |
| 1029 | |
| 1030 | /* Substitute with regex */ |
| 1031 | case 's': |
| 1032 | if (!do_subst_command(sed_cmd, &pattern_space)) |
| 1033 | break; |
| 1034 | substituted |= 1; |
| 1035 | |
| 1036 | /* handle p option */ |
| 1037 | if (sed_cmd->sub_p) |
| 1038 | sed_puts(pattern_space, last_gets_char); |
| 1039 | /* handle w option */ |
| 1040 | if (sed_cmd->sw_file) |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1041 | puts_maybe_newline( |
| 1042 | pattern_space, sed_cmd->sw_file, |
| 1043 | &sed_cmd->sw_last_char, last_gets_char); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1044 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1045 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1046 | /* Append line to linked list to be printed later */ |
| 1047 | case 'a': |
| 1048 | append(sed_cmd->string); |
| 1049 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1050 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1051 | /* Insert text before this line */ |
| 1052 | case 'i': |
| 1053 | sed_puts(sed_cmd->string, '\n'); |
| 1054 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1055 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1056 | /* Cut and paste text (replace) */ |
| 1057 | case 'c': |
| 1058 | /* Only triggers on last line of a matching range. */ |
| 1059 | if (!sed_cmd->in_match) |
Denys Vlasenko | 96a1833 | 2010-04-19 22:36:07 -0400 | [diff] [blame] | 1060 | sed_puts(sed_cmd->string, '\n'); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1061 | goto discard_line; |
| 1062 | |
| 1063 | /* Read file, append contents to output */ |
| 1064 | case 'r': |
| 1065 | { |
| 1066 | FILE *rfile; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1067 | rfile = fopen_for_read(sed_cmd->string); |
| 1068 | if (rfile) { |
| 1069 | char *line; |
| 1070 | |
| 1071 | while ((line = xmalloc_fgetline(rfile)) |
| 1072 | != NULL) |
| 1073 | append(line); |
| 1074 | xprint_and_close_file(rfile); |
| 1075 | } |
| 1076 | |
| 1077 | break; |
| 1078 | } |
| 1079 | |
| 1080 | /* Write pattern space to file. */ |
| 1081 | case 'w': |
| 1082 | puts_maybe_newline( |
| 1083 | pattern_space, sed_cmd->sw_file, |
| 1084 | &sed_cmd->sw_last_char, last_gets_char); |
| 1085 | break; |
| 1086 | |
| 1087 | /* Read next line from input */ |
| 1088 | case 'n': |
| 1089 | if (!G.be_quiet) |
| 1090 | sed_puts(pattern_space, last_gets_char); |
| 1091 | if (next_line) { |
| 1092 | free(pattern_space); |
| 1093 | pattern_space = next_line; |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1094 | last_gets_char = next_gets_char; |
| 1095 | next_line = get_next_line(&next_gets_char); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1096 | substituted = 0; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1097 | linenum++; |
| 1098 | break; |
| 1099 | } |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1100 | /* fall through */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1101 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1102 | /* Quit. End of script, end of input. */ |
| 1103 | case 'q': |
| 1104 | /* Exit the outer while loop */ |
| 1105 | free(next_line); |
| 1106 | next_line = NULL; |
| 1107 | goto discard_commands; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1108 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1109 | /* Append the next line to the current line */ |
| 1110 | case 'N': |
| 1111 | { |
| 1112 | int len; |
| 1113 | /* If no next line, jump to end of script and exit. */ |
| 1114 | if (next_line == NULL) { |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1115 | free(next_line); |
| 1116 | next_line = NULL; |
| 1117 | goto discard_line; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1118 | } |
Denys Vlasenko | bf5f99f | 2010-06-04 01:29:52 +0200 | [diff] [blame] | 1119 | /* Append next_line, read new next_line. */ |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1120 | len = strlen(pattern_space); |
Denys Vlasenko | 9037787 | 2010-01-08 09:07:50 +0100 | [diff] [blame] | 1121 | pattern_space = xrealloc(pattern_space, len + strlen(next_line) + 2); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1122 | pattern_space[len] = '\n'; |
| 1123 | strcpy(pattern_space + len+1, next_line); |
| 1124 | last_gets_char = next_gets_char; |
| 1125 | next_line = get_next_line(&next_gets_char); |
| 1126 | linenum++; |
| 1127 | break; |
| 1128 | } |
| 1129 | |
| 1130 | /* Test/branch if substitution occurred */ |
| 1131 | case 't': |
| 1132 | if (!substituted) break; |
| 1133 | substituted = 0; |
| 1134 | /* Fall through */ |
| 1135 | /* Test/branch if substitution didn't occur */ |
| 1136 | case 'T': |
| 1137 | if (substituted) break; |
| 1138 | /* Fall through */ |
| 1139 | /* Branch to label */ |
| 1140 | case 'b': |
| 1141 | if (!sed_cmd->string) goto discard_commands; |
| 1142 | else sed_cmd = branch_to(sed_cmd->string); |
| 1143 | break; |
| 1144 | /* Transliterate characters */ |
| 1145 | case 'y': |
| 1146 | { |
| 1147 | int i, j; |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1148 | for (i = 0; pattern_space[i]; i++) { |
| 1149 | for (j = 0; sed_cmd->string[j]; j += 2) { |
| 1150 | if (pattern_space[i] == sed_cmd->string[j]) { |
| 1151 | pattern_space[i] = sed_cmd->string[j + 1]; |
| 1152 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1153 | } |
| 1154 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1155 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1156 | |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1157 | break; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1158 | } |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1159 | case 'g': /* Replace pattern space with hold space */ |
| 1160 | free(pattern_space); |
| 1161 | pattern_space = xstrdup(G.hold_space ? G.hold_space : ""); |
| 1162 | break; |
| 1163 | case 'G': /* Append newline and hold space to pattern space */ |
| 1164 | { |
| 1165 | int pattern_space_size = 2; |
| 1166 | int hold_space_size = 0; |
| 1167 | |
| 1168 | if (pattern_space) |
| 1169 | pattern_space_size += strlen(pattern_space); |
| 1170 | if (G.hold_space) |
| 1171 | hold_space_size = strlen(G.hold_space); |
| 1172 | pattern_space = xrealloc(pattern_space, |
| 1173 | pattern_space_size + hold_space_size); |
| 1174 | if (pattern_space_size == 2) |
| 1175 | pattern_space[0] = 0; |
| 1176 | strcat(pattern_space, "\n"); |
| 1177 | if (G.hold_space) |
| 1178 | strcat(pattern_space, G.hold_space); |
| 1179 | last_gets_char = '\n'; |
| 1180 | |
| 1181 | break; |
| 1182 | } |
| 1183 | case 'h': /* Replace hold space with pattern space */ |
| 1184 | free(G.hold_space); |
| 1185 | G.hold_space = xstrdup(pattern_space); |
| 1186 | break; |
| 1187 | case 'H': /* Append newline and pattern space to hold space */ |
| 1188 | { |
| 1189 | int hold_space_size = 2; |
| 1190 | int pattern_space_size = 0; |
| 1191 | |
| 1192 | if (G.hold_space) |
| 1193 | hold_space_size += strlen(G.hold_space); |
| 1194 | if (pattern_space) |
| 1195 | pattern_space_size = strlen(pattern_space); |
| 1196 | G.hold_space = xrealloc(G.hold_space, |
| 1197 | hold_space_size + pattern_space_size); |
| 1198 | |
| 1199 | if (hold_space_size == 2) |
| 1200 | *G.hold_space = 0; |
| 1201 | strcat(G.hold_space, "\n"); |
| 1202 | if (pattern_space) |
| 1203 | strcat(G.hold_space, pattern_space); |
| 1204 | |
| 1205 | break; |
| 1206 | } |
| 1207 | case 'x': /* Exchange hold and pattern space */ |
| 1208 | { |
| 1209 | char *tmp = pattern_space; |
Denys Vlasenko | 90a9904 | 2009-09-06 02:36:23 +0200 | [diff] [blame] | 1210 | pattern_space = G.hold_space ? G.hold_space : xzalloc(1); |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1211 | last_gets_char = '\n'; |
| 1212 | G.hold_space = tmp; |
| 1213 | break; |
| 1214 | } |
| 1215 | } /* switch */ |
| 1216 | } /* for each cmd */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1217 | |
| 1218 | /* |
Denys Vlasenko | 8bca3e2 | 2009-06-30 19:19:37 +0200 | [diff] [blame] | 1219 | * Exit point from sedding... |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1220 | */ |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1221 | discard_commands: |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1222 | /* we will print the line unless we were told to be quiet ('-n') |
| 1223 | or if the line was suppressed (ala 'd'elete) */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1224 | if (!G.be_quiet) |
Denis Vlasenko | fe7a9f1 | 2007-01-29 14:31:47 +0000 | [diff] [blame] | 1225 | sed_puts(pattern_space, last_gets_char); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1226 | |
| 1227 | /* Delete and such jump here. */ |
Denis Vlasenko | 826c85f | 2007-01-28 23:26:15 +0000 | [diff] [blame] | 1228 | discard_line: |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1229 | flush_append(); |
| 1230 | free(pattern_space); |
| 1231 | |
| 1232 | goto again; |
Erik Andersen | 1266a13 | 1999-12-29 22:19:46 +0000 | [diff] [blame] | 1233 | } |
| 1234 | |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1235 | /* It is possible to have a command line argument with embedded |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1236 | * newlines. This counts as multiple command lines. |
| 1237 | * However, newline can be escaped: 's/e/z\<newline>z/' |
| 1238 | * We check for this. |
| 1239 | */ |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1240 | |
| 1241 | static void add_cmd_block(char *cmdstr) |
| 1242 | { |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1243 | char *sv, *eol; |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1244 | |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1245 | cmdstr = sv = xstrdup(cmdstr); |
| 1246 | do { |
| 1247 | eol = strchr(cmdstr, '\n'); |
| 1248 | next: |
| 1249 | if (eol) { |
| 1250 | /* Count preceding slashes */ |
| 1251 | int slashes = 0; |
| 1252 | char *sl = eol; |
| 1253 | |
| 1254 | while (sl != cmdstr && *--sl == '\\') |
| 1255 | slashes++; |
| 1256 | /* Odd number of preceding slashes - newline is escaped */ |
| 1257 | if (slashes & 1) { |
Denis Vlasenko | 0f293b9 | 2008-07-22 20:16:55 +0000 | [diff] [blame] | 1258 | overlapping_strcpy(eol - 1, eol); |
Denis Vlasenko | 4b0bb9e | 2007-03-16 23:36:58 +0000 | [diff] [blame] | 1259 | eol = strchr(eol, '\n'); |
| 1260 | goto next; |
| 1261 | } |
| 1262 | *eol = '\0'; |
| 1263 | } |
| 1264 | add_cmd(cmdstr); |
| 1265 | cmdstr = eol + 1; |
| 1266 | } while (eol); |
| 1267 | free(sv); |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1268 | } |
| 1269 | |
Denis Vlasenko | 9b49a5e | 2007-10-11 10:05:36 +0000 | [diff] [blame] | 1270 | int sed_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
Denis Vlasenko | a60f84e | 2008-07-05 09:18:54 +0000 | [diff] [blame] | 1271 | int sed_main(int argc UNUSED_PARAM, char **argv) |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1272 | { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1273 | enum { |
| 1274 | OPT_in_place = 1 << 0, |
| 1275 | }; |
Denis Vlasenko | 67b23e6 | 2006-10-03 21:00:06 +0000 | [diff] [blame] | 1276 | unsigned opt; |
Denis Vlasenko | b97c984 | 2006-10-01 21:05:12 +0000 | [diff] [blame] | 1277 | llist_t *opt_e, *opt_f; |
| 1278 | int status = EXIT_SUCCESS; |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1279 | |
Denis Vlasenko | 74324c8 | 2007-06-04 10:16:52 +0000 | [diff] [blame] | 1280 | INIT_G(); |
Rob Landley | e3f5a3f | 2006-05-09 03:53:55 +0000 | [diff] [blame] | 1281 | |
Mark Whitley | 858c1ad | 2000-07-11 21:38:47 +0000 | [diff] [blame] | 1282 | /* destroy command strings on exit */ |
Rob Landley | 0b65628 | 2006-05-08 22:17:23 +0000 | [diff] [blame] | 1283 | if (ENABLE_FEATURE_CLEAN_UP) atexit(sed_free_and_close_stuff); |
Mark Whitley | 858c1ad | 2000-07-11 21:38:47 +0000 | [diff] [blame] | 1284 | |
Rob Landley | fae1dc8 | 2005-11-20 07:44:35 +0000 | [diff] [blame] | 1285 | /* Lie to autoconf when it starts asking stupid questions. */ |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 1286 | if (argv[1] && !strcmp(argv[1], "--version")) { |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1287 | puts("This is not GNU sed version 4.0"); |
| 1288 | return 0; |
Eric Andersen | c06f568 | 2004-02-04 10:57:46 +0000 | [diff] [blame] | 1289 | } |
Eric Andersen | c06f568 | 2004-02-04 10:57:46 +0000 | [diff] [blame] | 1290 | |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1291 | /* do normal option parsing */ |
Denis Vlasenko | b97c984 | 2006-10-01 21:05:12 +0000 | [diff] [blame] | 1292 | opt_e = opt_f = NULL; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1293 | opt_complementary = "e::f::" /* can occur multiple times */ |
| 1294 | "nn"; /* count -n */ |
Denis Vlasenko | fe7cd64 | 2007-08-18 15:32:12 +0000 | [diff] [blame] | 1295 | opt = getopt32(argv, "irne:f:", &opt_e, &opt_f, |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1296 | &G.be_quiet); /* counter for -n */ |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 1297 | //argc -= optind; |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1298 | argv += optind; |
| 1299 | if (opt & OPT_in_place) { // -i |
Denis Vlasenko | 750fc6d | 2006-09-22 08:56:03 +0000 | [diff] [blame] | 1300 | atexit(cleanup_outname); |
| 1301 | } |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1302 | if (opt & 0x2) G.regex_type |= REG_EXTENDED; // -r |
| 1303 | //if (opt & 0x4) G.be_quiet++; // -n |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 1304 | while (opt_e) { // -e |
Denis Vlasenko | d50dda8 | 2008-06-15 05:40:56 +0000 | [diff] [blame] | 1305 | add_cmd_block(llist_pop(&opt_e)); |
Denis Vlasenko | 750fc6d | 2006-09-22 08:56:03 +0000 | [diff] [blame] | 1306 | } |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 1307 | while (opt_f) { // -f |
| 1308 | char *line; |
| 1309 | FILE *cmdfile; |
Denis Vlasenko | 5415c85 | 2008-07-21 23:05:26 +0000 | [diff] [blame] | 1310 | cmdfile = xfopen_for_read(llist_pop(&opt_f)); |
Denis Vlasenko | 8ee649a | 2008-03-26 20:04:27 +0000 | [diff] [blame] | 1311 | while ((line = xmalloc_fgetline(cmdfile)) != NULL) { |
Denis Vlasenko | 945bd3d | 2007-04-12 21:20:25 +0000 | [diff] [blame] | 1312 | add_cmd(line); |
| 1313 | free(line); |
| 1314 | } |
| 1315 | fclose(cmdfile); |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1316 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1317 | /* if we didn't get a pattern from -e or -f, use argv[0] */ |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 1318 | if (!(opt & 0x18)) { |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 1319 | if (!*argv) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 1320 | bb_show_usage(); |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1321 | add_cmd_block(*argv++); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1322 | } |
Glenn L McGrath | 42c2573 | 2003-10-04 05:27:56 +0000 | [diff] [blame] | 1323 | /* Flush any unfinished commands. */ |
| 1324 | add_cmd(""); |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1325 | |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 1326 | /* By default, we write to stdout */ |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1327 | G.nonstdout = stdout; |
Rob Landley | 53302f8 | 2004-02-18 09:54:15 +0000 | [diff] [blame] | 1328 | |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1329 | /* argv[0..(argc-1)] should be names of file to process. If no |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1330 | * files were specified or '-' was specified, take input from stdin. |
| 1331 | * Otherwise, we process all the files specified. */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1332 | if (argv[0] == NULL) { |
| 1333 | if (opt & OPT_in_place) |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 1334 | bb_error_msg_and_die(bb_msg_requires_arg, "-i"); |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 1335 | add_input_file(stdin); |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 1336 | } else { |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1337 | int i; |
| 1338 | FILE *file; |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 1339 | |
Denis Vlasenko | 62a90cd | 2008-03-17 09:07:36 +0000 | [diff] [blame] | 1340 | for (i = 0; argv[i]; i++) { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1341 | struct stat statbuf; |
| 1342 | int nonstdoutfd; |
| 1343 | |
Denis Vlasenko | 9f73944 | 2006-12-16 23:49:13 +0000 | [diff] [blame] | 1344 | if (LONE_DASH(argv[i]) && !(opt & OPT_in_place)) { |
Rob Landley | dcc2866 | 2004-11-25 07:21:47 +0000 | [diff] [blame] | 1345 | add_input_file(stdin); |
| 1346 | process_files(); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1347 | continue; |
Glenn L McGrath | 8aac05b | 2003-09-14 04:06:12 +0000 | [diff] [blame] | 1348 | } |
Denis Vlasenko | ddec5af | 2006-10-26 23:25:17 +0000 | [diff] [blame] | 1349 | file = fopen_or_warn(argv[i], "r"); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1350 | if (!file) { |
| 1351 | status = EXIT_FAILURE; |
| 1352 | continue; |
| 1353 | } |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1354 | if (!(opt & OPT_in_place)) { |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1355 | add_input_file(file); |
| 1356 | continue; |
| 1357 | } |
| 1358 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1359 | G.outname = xasprintf("%sXXXXXX", argv[i]); |
| 1360 | nonstdoutfd = mkstemp(G.outname); |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 1361 | if (-1 == nonstdoutfd) |
Denys Vlasenko | 6331cf0 | 2009-11-13 09:08:27 +0100 | [diff] [blame] | 1362 | bb_perror_msg_and_die("can't create temp file %s", G.outname); |
Denys Vlasenko | a7ccdee | 2009-11-15 23:28:11 +0100 | [diff] [blame] | 1363 | G.nonstdout = xfdopen_for_write(nonstdoutfd); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1364 | |
Denys Vlasenko | 5799248 | 2009-11-13 09:09:07 +0100 | [diff] [blame] | 1365 | /* Set permissions/owner of output file */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1366 | fstat(fileno(file), &statbuf); |
Denys Vlasenko | 3b727cc | 2010-06-18 02:12:56 +0200 | [diff] [blame] | 1367 | /* chmod'ing AFTER chown would preserve suid/sgid bits, |
| 1368 | * but GNU sed 4.2.1 does not preserve them either */ |
Denis Vlasenko | 2f8f71b | 2006-12-10 02:09:12 +0000 | [diff] [blame] | 1369 | fchmod(nonstdoutfd, statbuf.st_mode); |
Denys Vlasenko | 5799248 | 2009-11-13 09:09:07 +0100 | [diff] [blame] | 1370 | fchown(nonstdoutfd, statbuf.st_uid, statbuf.st_gid); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1371 | add_input_file(file); |
| 1372 | process_files(); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1373 | fclose(G.nonstdout); |
Denis Vlasenko | d18a3a2 | 2006-10-25 12:46:03 +0000 | [diff] [blame] | 1374 | |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1375 | G.nonstdout = stdout; |
Denis Vlasenko | 8077850 | 2006-10-25 12:46:46 +0000 | [diff] [blame] | 1376 | /* unlink(argv[i]); */ |
Denis Vlasenko | cb448fe | 2008-02-17 14:28:53 +0000 | [diff] [blame] | 1377 | xrename(G.outname, argv[i]); |
Denis Vlasenko | 17a1526 | 2007-03-26 20:48:46 +0000 | [diff] [blame] | 1378 | free(G.outname); |
Denis Vlasenko | 4027664 | 2007-11-13 16:48:10 +0000 | [diff] [blame] | 1379 | G.outname = NULL; |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1380 | } |
Denys Vlasenko | eb08b6e | 2010-06-19 17:51:06 +0200 | [diff] [blame] | 1381 | /* Here, to handle "sed 'cmds' nonexistent_file" case we did: |
| 1382 | * if (G.current_input_file >= G.input_file_count) |
| 1383 | * return status; |
| 1384 | * but it's not needed since process_files() works correctly |
| 1385 | * in this case too. */ |
Mark Whitley | 6315ce6 | 2000-07-10 22:55:51 +0000 | [diff] [blame] | 1386 | } |
Denys Vlasenko | eb08b6e | 2010-06-19 17:51:06 +0200 | [diff] [blame] | 1387 | process_files(); |
Glenn L McGrath | 8d6395d | 2003-04-08 11:56:11 +0000 | [diff] [blame] | 1388 | |
Matt Kraai | a5f09c6 | 2001-11-12 16:44:55 +0000 | [diff] [blame] | 1389 | return status; |
Eric Andersen | 6b6b3f6 | 1999-10-28 16:06:25 +0000 | [diff] [blame] | 1390 | } |